add CHROME_TIMEOUT args

Signed-off-by: ふぁ <[email protected]>

archivebox/config.py

@@ -139,6 +139,7 @@ CONFIG_SCHEMA: Dict[str, ConfigDefaultDict] = {
         'COOKIES_FILE':             {'type': str,   'default': None},
         'CHROME_USER_DATA_DIR':     {'type': str,   'default': None},
 
+        'CHROME_TIMEOUT':           {'type': int,  'default': 0},
         'CHROME_HEADLESS':          {'type': bool,  'default': True},
         'CHROME_SANDBOX':           {'type': bool,  'default': lambda c: not c['IN_DOCKER']},
         'YOUTUBEDL_ARGS':           {'type': list,  'default': lambda c: [
@@ -981,6 +982,7 @@ def get_chrome_info(config: ConfigDict) -> ConfigValue:
         'RESOLUTION': config['RESOLUTION'],
         'CHECK_SSL_VALIDITY': config['CHECK_SSL_VALIDITY'],
         'CHROME_BINARY': bin_path(config['CHROME_BINARY']),
+        'CHROME_TIMEOUT':config['CHROME_TIMEOUT'],
         'CHROME_HEADLESS': config['CHROME_HEADLESS'],
         'CHROME_SANDBOX': config['CHROME_SANDBOX'],
         'CHROME_USER_AGENT': config['CHROME_USER_AGENT'],

archivebox/config_stubs.py

@@ -74,6 +74,7 @@ class ConfigDict(BaseConfig, total=False):
     CHROME_USER_AGENT: str
     COOKIES_FILE: Union[str, Path, None]
     CHROME_USER_DATA_DIR: Union[str, Path, None]
+    CHROME_TIMEOUT: int
     CHROME_HEADLESS: bool
     CHROME_SANDBOX: bool
 

archivebox/extractors/dom.py

@@ -39,7 +39,7 @@ def save_dom(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEOUT) ->
     output: ArchiveOutput = 'output.html'
     output_path = out_dir / output
     cmd = [
-        *chrome_args(TIMEOUT=timeout),
+        *chrome_args(),
         '--dump-dom',
         link.url
     ]

archivebox/extractors/pdf.py

@@ -37,7 +37,7 @@ def save_pdf(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEOUT) ->
     out_dir = out_dir or Path(link.link_dir)
     output: ArchiveOutput = 'output.pdf'
     cmd = [
-        *chrome_args(TIMEOUT=timeout),
+        *chrome_args(),
         '--print-to-pdf',
         link.url,
     ]

archivebox/extractors/screenshot.py

@@ -37,7 +37,7 @@ def save_screenshot(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEO
     out_dir = out_dir or Path(link.link_dir)
     output: ArchiveOutput = 'screenshot.png'
     cmd = [
-        *chrome_args(TIMEOUT=timeout),
+        *chrome_args(),
         '--screenshot',
         link.url,
     ]

archivebox/extractors/singlefile.py

@@ -42,7 +42,7 @@ def save_singlefile(link: Link, out_dir: Optional[Path]=None, timeout: int=TIMEO
     out_dir = out_dir or Path(link.link_dir)
     output = "singlefile.html"
 
-    browser_args = chrome_args(TIMEOUT=0)
+    browser_args = chrome_args(CHROME_TIMEOUT=0)
 
     # SingleFile CLI Docs: https://github.com/gildas-lormeau/SingleFile/tree/master/cli
     browser_args = '--browser-args={}'.format(json.dumps(browser_args[1:]))

archivebox/util.py

@@ -260,8 +260,8 @@ def chrome_args(**options) -> List[str]:
     if options['RESOLUTION']:
         cmd_args += ('--window-size={}'.format(options['RESOLUTION']),)
 
-    #if options['TIMEOUT']:
-    #    cmd_args += ('--timeout={}'.format(options['TIMEOUT'] * 1000),)
+    if options['CHROME_TIMEOUT']:
+       cmd_args += ('--timeout={}'.format(options['CHROME_TIMEOUT'] * 1000),)
 
     if options['CHROME_USER_DATA_DIR']:
         cmd_args.append('--user-data-dir={}'.format(options['CHROME_USER_DATA_DIR']))